Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
NBER WORKING PAPER SERIES
INITIAL COIN OFFERINGS AND THE VALUE OF CRYPTO TOKENS
Christian CataliniJoshua S. Gans
Working Paper 24418http://www.nber.org/papers/w24418
NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue
Cambridge, MA 02138March 2018, Revised March 2019
We thank Jane Wu and participants at the University of Chicago, University of Michigan and University of Toronto for useful comments. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.
At least one co-author has disclosed a financial relationship of potential relevance for this research. Further information is available online at http://www.nber.org/papers/w24418.ack
NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies official NBER publications.
© 2018 by Christian Catalini and Joshua S. Gans. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source.
Initial Coin Offerings and the Value of Crypto Tokens Christian Catalini and Joshua S. GansNBER Working Paper No. 24418March 2018, Revised March 2019JEL No. E42,L12,L26
ABSTRACT
This paper explores how entrepreneurs can use initial coin offerings — whereby they issue crypto tokens and commit to only accept those tokens as payment for their products — to fund venture start-up costs. We show that the ICO mechanism allows entrepreneurs to generate buyer competition for the token, giving it value. We also find that venture returns are independent of any committed growth in the supply of tokens over time, but that initial funds raised are maximized by setting that growth to zero to encourage saving by early participants. Nonetheless, since the value of the tokens depends on a single period of demand, the ability to raise funds is more limited than in traditional equity finance. Furthermore, a lack of commitment in monetary policy undermines saving behavior, hence the cost of using tokens to fund start-up costs is inflexibility in future capital raises. Crypto tokens can also facilitate coordination among stakeholders within digital ecosystems when network effects are present.
Christian CataliniMIT Sloan School of Management100 Main Street, E62-480Cambridge, MA 02142and [email protected]
Joshua S. GansRotman School of ManagementUniversity of Toronto105 St. George StreetToronto ON M5S 3E6CANADAand [email protected]
1. Introduction
Initial coin offerings (ICOs) have emerged as a novel mechanism for financing
entrepreneurial ventures. Through an ICO, a venture offers a stock of specialized crypto tokens
for sale with the promise that those tokens will operate as the only medium of exchange when
accessing the venture’s future products. The sale of tokens provides capital to fund initial
development, although no commitment is made as to the price of future products (in tokens or
otherwise). Since 2017, blockchain startups have raised over $7B through initial coin offerings 2 3
compared to $1B through traditional venture capital flowing into the space (Catalini et al., 2017).
Approximately one third of all ICO funding went to US-based teams, and more than 200 ICOs
raised above $10M. 4
While the idea of issuing firm-specific tokens dates back to de Bono (1994), the recent
spike in such activity follows the invention of Bitcoin by Nakamoto (2008), and the development
of cryptocurrencies with additional programming capabilities such as Ethereum. Using these
platforms, a venture can fund its development with extremely low frictions through the issuance
and auctioning off of dedicated crypto tokens. This is the result of blockchain technology
lowering both the cost of verifying transaction attributes — which allows for the self-custody of
In this respect, token sales have a pre-sale aspect similar to crowdfunding, but differ in that there is no pre-sale 2
price commitment to token holders (cf: Agrawal, Catalini and Goldfarb, 2013).
To place this number into perspective, crowdfunding platform Kickstarter, over the course of 9 years, allocated a 3
total of $3.5B to entrepreneurial and artistic projects. Equity crowdfunding platform AngelList, through its syndicated model, facilitated approximately $700M in online, early stage equity investments since 2013.
Among recent offerings, Tezos raised $232M for developing a smart contracts and decentralized governance 4
platform; Filecoin $205M from over 2,100 accredited investors to deploy a decentralized file storage network; Kin $98M to build a decentralized social network and communication platform; Blockstack $52M towards a decentralized browser, identity and application ecosystem; and BAT $35M to develop a blockchain-based digital advertising ecosystem.
!2
digital assets — and the cost of coordinating economic activity over the internet (Catalini and
Gans, 2016).
This paper provides an economic analysis of the ICO funding mechanism and how it both
differs and relates to traditional equity financing. In traditional equity financing, a venture issues 5
shares that are claims on the discounted sum of future profits of the venture. Equity value is
determined by the trading price for those shares. By contrast, when specialized tokens are issued,
they are not claims on future profits, but are instead only related, at best, to the flow of future
revenues. This is because it is the flow of revenues that will determine the exchange rate of
tokens against fiat-currencies in the future. We argue that it is this fundamental insight that is
required to understand ICOs as a funding mechanism, and the conditions under which they may
be an effective source of capital for entrepreneurs.
What is unclear about ICOs is how tokens can have any value given that the venture
issuing them controls both their technological evolution as well as their monetary policy. In a
basic implementation, investors only have the right to spend tokens to purchase products from
the venture. At the time of fundraising, neither the venture nor the investors can predict the future
demand for the tokens, and the venture cannot credibly commit to a future token-denominated
price, as committing to the ‘wrong’ price would lead to zero value to the venture. Moreover, if
the venture incurs additional costs to deliver their products and operate the ecosystem around the
token, then it would have an incentive to set the token price high to limit demand and economize
on costs.
To our knowledge, it is the first analysis of ICOs as an alternative source of entrepreneurial financing. Other recent 5
and contemporaneous papers have focused on ICOs as a coordinating mechanism when there are network effects — something we also remark upon later in this paper. See Cong, Li and Wang (2018), Fisch (2018), Li and Mann (2018), and Bakos and Halaburda (2018).
!3
In this paper, we examine how a token that can only be used to transact on a specific
platform can have value in the absence of additional rights over the venture itself, its governance,
or its future profits (i.e. we do not study crypto securities that resemble the rights of traditional
equity arrangements). To build a model of crypto tokens and understand how they can have value
and be used to raise capital, we abstract away from cases where entrepreneurs might fraudulently
issue tokens with no intention of creating a venture. We instead focus on situations where a
venture — if viable — will be created, and fraudulent entrepreneurs or those without the ability
to execute on their promises are unlikely to be funded (i.e. a market for curation of token
offerings has developed). Even in the absence of fraud and incompetence, it is not obvious how
tokens precisely have value without additional rights on the future profits of the venture.
Our main findings are as follows. First, after an ICO, when setting their token-denominated
prices, ventures use a simple ‘divide the money’ rule that in equilibrium leads to the same dollar-
denominated price consumers would pay in the absence of tokens. As a result ventures have no
incentive to commit ex ante to a specific price. Second, there is a fundamental trade-off between
price stability of token-denominated prices and the amount that can be raised through an ICO: to
raise more upfront, a venture has to maximize the use of tokens as a store of value (i.e. it has to
encourage holding behavior through future appreciation), but this undermines the ability to use
the tokens as a stable medium of exchange. Third, we provide conditions under which the
venture needs to retain partial ownership of its tokens to mitigate incentives to later raise token-
denominated prices and under-supply their product. We also highlight the importance of two key 6
commitments by the venture: limiting the token supply, and enforcing the use of the token as the
This has implications for securities regulation which examines venture control over an asset to determine whether 6
it should be subject to regulation or not.
!4
only medium of exchange when transacting with the venture. Last, we show that ventures relying
on ICOs will face constraints in raising follow-on capital through the same mechanism, and that
perfectly viable ventures which could have raised sufficient capital through traditional sources,
may fail to do so through ICOs. This issue is particularly severe when the venture is long-lived,
and is consistent with the rise of hybrid arrangements where ventures raise a traditional equity
round before issuing tokens to the public or to accredited investors. 7
There has been a burgeoning literature studying cryptocurrencies and their use as a novel
source of early-stage capital. While several empirical papers have examined different aspects of
ICOs (Catalini, Boslego, and Zhang, 2017; Benedetti and Kostovetsky, 2018; Howell, Niessner
and Yermack, 2018; Momtaz, 2018), these are yet to draw on theory. On the theoretical side,
some papers examine the role of cryptocurrencies in the adoption of platforms (Cong, Li and
Wang, 2018; Fisch, 2018; Li and Mann, 2018; Bakos and Halaburda, 2018), and others have
examined the conditions under which blockchain technology can help enforce cryptocurrency
commitments (Halaburda and Sarvary, 2016; Biais, Bisiere, Bouvard and Casamatta, 2017;
Budish, 2018). While we discuss platform implications towards the end of the paper, our main
focus is to explore the more general idea of using tokens as venture-specific media of
fundraising and exchange. We also abstract away from issues of blockchain stability and instead
rely on characteristics of the technology to motivate the commitments — specifically to the
supply of tokens — that are now feasible. Finally, and most closely related to this paper, there is
While the returns to the platform (which often constitutes an open source software protocol and can be considered 7
as “shared infrastructure” among all participants within an ecosystem) can be appropriated by all early stage investors through the direct appreciation of the token, the returns to the broader set of products the venture may create over time (e.g. new applications on top of the shared protocol) only accrue to equity holders. Because of the inherent uncertainty about which component will be more valuable in the long run — between the underlying protocol and the additional products a venture or third-parties may develop on top of it — venture capital firms have started writing hybrid contracts where they receive both tokens and equity in exchange for funding.
!5
research that examines the financial market characteristics of crypto tokens (Sockin and Xiong,
2018; Chod and Lyandres, 2018; Canidio, 2018). These papers explore different issues including
the role of information asymmetry between investors and entrepreneurs (Chod and Lyandres,
2018) or between platform users (Sockin and Xiong, 2018), and the ability to sell tokens without
developing a service (Canidio, 2018). Our model is, in a sense, more baseline in that it focuses
on the determinants of token value in the absence of these factors. Thus, this body of theoretical
research can be regarded as complementary to this paper.
We proceed by building a simple model starting from the classical situation of an
entrepreneur who needs to raise equity financing from a venture capitalist because of financing
constraints (Section 2). We use this template for equity financing as our starting point, and then
benchmark it against ICOs (Section 3). In Section 4, we study what happens when the venture
faces a standard demand function, when tokens can be retired, when there are ongoing
operational costs, and when additional third-parties can provide services using the tokens
launched by the venture. Section 5 looks at imperfect commitment, and in particular at what
happens when the venture cannot commit to a pre-determined monetary policy or to making the
tokens the only medium of exchange within its ecosystem. Section 6 explores ICOs in the
presence of network effects. The last section concludes.
2. Model Set-Up
We model an entrepreneur who faces an upfront cost, C, of creating a venture. Once this
cost is sunk, the venture can deploy its product solution and start operating in the following
period. There are three time periods, ! , and all agents in the model have a common t ∈{0,1,2}
!6
discount factor, ! . Revenues can only be generated starting one period after the venture is
launched. Thus, our focus is on how a liquidity constrained entrepreneur raises pre-revenue funds
to finance the upfront development costs of their venture.
There are a number of (small) buyers each placing the same value, ! , on product
quality. Let nt — the number of buyers in period t — be a measure of demand where we assume that 8
n0 = 0. Product quality, q, and period 1 demand, n1, are initially unknown and distributed according
to a cdf, . We assume that the number of buyers in the second period, n2, is fully
determined from the realization of n1 according to , where gn (demand growth) is
known from the outset.
Once C is sunk (and the venture is created), all buyers and the entrepreneur learn q and n1.
Therefore, when referring to the quantity of demand, in what follows we let . Finally, if et
is the token to dollar exchange rate in period t, and the venture’s token denominated price is pt,
then the dollar denominated price to buyers is ! .
No financing constraint
As a benchmark, suppose that the entrepreneur does not face a financing constraint and has C
in funds. When the venture is launched, q becomes common knowledge and so the entrepreneur
sets Pt = q. Given this, the venture will be launched if:
! .
δ ∈[0,1]
q ∈[0,1]
F(q,n1)
n2 = (1+ gn )n1
n = n1
Pt = et pt
δ (E[nP1]+δ (1+ gn )E[nP2]) = δ (1+δ (1+ gn ))E[nq]≥ C
We use the term ‘product’ here because there is nothing in the model below that restricts the outcome to digital 8
services per se even though all of the use cases we have seen to date are digital services. Of course, this definition of product is an economic one that encompasses digital services as a special case.
!7
Equity financing
Suppose now there is a competitive venture capital market that can provide C to liquidity
constrained entrepreneurs. How much equity (i.e., share of profits, ! ) will an entrepreneur need
to cede in order to raise C ? The minimum equity an investor will accept to finance the venture is:
!
Thus, the entrepreneur’s expected return is:
! .
As in the no financing constraint case, whether a venture proceeds or not depends on whether the
discounted expected revenues are greater than the venture start-up costs.
3. Initial Coin Offerings
In an ICO, entrepreneurs specify an amount they aim to raise. That amount is usually a cap,
and the entrepreneurs may retain a share of the tokens offered and be exposed to fluctuations in
the value of their crypto token. The timeline is as follows:
1. ICO stage • The entrepreneur sets the quantity of tokens m0; the minimum price each token will be
issued at e (in exchange for dollars); the share of tokens the entrepreneur will retain a,
and whether the ICO is made contingent on (1 − a)m0 tokens being purchased ex ante.
The entrepreneur also specifies the tokens available in the following two periods (m1
and m2).
• The entrepreneur then auctions the tokens (in either a multi-unit English auction or
second price auction). Other agents decide whether to purchase tokens or not.
1 − α
1−α * = Cδ (1+δ (1+ gn ))E[nq]
α *δ (1+δ (1+ gn ))E[nq] = δ (1+δ (1+ gn ))E[nq]−C
!8
• If the total tokens purchased exceed the minimum threshold, the entrepreneur proceeds
with the venture, otherwise all contributions are returned, the venture does not launch
and the game ends. 9
2. Market stage
• One period after the venture is created (through the sinking of cost C), product
quality is revealed to all uninformed agents.
• The entrepreneur launches the venture with tokens being the only accepted medium of
exchange for its products.
• Buyers trade tokens at a new market determined exchange rate.
• Payoffs and profits are realized.
Following Athey et. al. (2016), a dynamic price equilibrium requires the following:
(a) (Agent optimization) Each buyer chooses to purchase the product in period t if ! . An agent chooses to purchase tokens at the end of a period if ! . The
venture sets a price in each period t to maximize ! , the dollar value of revenue, where ! (that is, the number of units purchased). The choice
of ! maximizes the expected net present discounted value of venture profits.
(b) (Market clearing) The market for tokens clears at the maximum exchange rate such that the demand for tokens is less than supply.
(c) (Rational expectations) Agents’ expectation of the next period’s exchange rate are correct.
Market stage
To examine the ICO process, we work backwards and start by examining the final market
stage. Suppose that mt tokens are available at time t (i.e. total issued tokens at t minus those being
held between t and t + 1). The following proposition characterizes the (token-denominated)
price, pt, set by the entrepreneurs given their knowledge of q.
et pt ≤ q et ≤ δet+1
et pt D(et pt )D(et pt ) = # Iet pt≤q
(m1,m2 ,a)
This is similar to the provision point mechanism adopted on crowdfunding platforms.9
!9
Proposition 1. A dynamic price equilibrium exists where ! and all consumers with a positive value for the product purchase it during the market stage.
The proof is as follows. If product quality is revealed to be q, then the individual demand for
tokens in dollars will be ptet so long as . In equilibrium, the exchange rate, et, will be
set by market clearing. The exchange rate depends on whether ntpt (token demand) is less than,
equal to or greater than mt (token supply):
• If pt < mt/nt, then products can be purchased without using all of the token supply. In this
case, et will tend towards 0 in order to clear the token market. This will give the venture no
revenue in dollar terms.
• If pt > mt/nt instead, tokens will be scarce and total token demand will be less than ntq as
some customers are excluded.
• Last, if pt = mt/nt, then the (aggregate) dollar demand for tokens will be exactly ntq, making
this the optimal price choice for the product. In this scenario, the exchange rate will be
determined by: q/et = mt/nt, which occurs when e(mt) = ntq/mt (i.e. when all token holdings
are used by consumers to purchase the venture’s products).
It is useful to reflect on what this means from a pricing strategy perspective. Without
tokens, the venture would price based on expected willingness to pay. With tokens, it does not
have control over the exchange rate and so cannot directly price in the same way. Instead, what it
does is target the number of product units it wants to sell which, in this model, is the same as the
number of expected consumers in each period. It then sets a divide the money price which
divides the available supply of tokens equally among the expected number of consumers.
Consumers then bid for tokens if they they wish to purchase the product, leading to the exchange
pt* = mt / nt
ptet ≤ ntq
!10
rate reflecting their willingness to pay. Importantly, this pricing strategy does not require the 10
venture to have direct knowledge of (or even expectations about) consumers’ willingness to pay:
the scarcity of tokens induced by the divide the money price reveals consumer value through
buyer competition. It is the ability to choose price in this manner that gives tokens value post-
ICO, and explains why tokens can have value even in the absence of the additional rights
typically associated with traditional securities.
Note that the equilibrium described in Proposition 1 is far from unique. At the beginning of
a market period, the venture may not have any tokens. In that case, when it sets pt, it will receive
tokens back as payment but there is no guarantee that those tokens have any value. In fact, given
that at the end of period 2 those tokens have no value, the venture could even set pt so high there
is no demand for its products. Consequently, there are multiple possible equilibria but we focus
on the one described in Proposition 1 because it maximizes the ongoing value of the tokens.
However, in Section 4, we consider a situation in which the venture has ongoing costs and,
therefore, has an incentive to set pt so high that there is no demand for its products. Nonetheless,
we demonstrate that the pricing outcome in Proposition 1 is an equilibrium in that case too.
The value of price commitments
When prices are set ex post, expected revenues for the venture are ! .
Since this plays an important role in determining the ex ante value of crypto tokens, it is useful to
examine whether an earlier price commitment at the time of the ICO (i.e., in period 0 before
uncertainty is resolved) can improve on this outcome.
(1+δ (1+ gn ))E[nq]
As the venture is receiving payment for those tokens, the exchange rate — so long as it is stable — will give them 10
dollar payments based on willingness to pay.
!11
Suppose that, prior to sinking C, the venture commits to p1 and p2. If pt < mt/nt, then et = 0
and the venture’s expected revenues in period t would be 0. If pt > mt/nt, then et = q/pt. This is
also true for pt = mt/nt. Thus, period 1 expected revenues would be . It is
easy to see that this is less than ! . An analogous argument applies to period 2. Intuitively,
the divide the money price induces the dollar-denominated monopoly revenue for each period.
Because of the uncertainty, it is impossible to commit ex ante to this price and there is always a
range of outcomes where either no dollar denominated revenues are generated, or a reduced level
of revenues arises as not all buyers are able to purchase the product with the available token
supply. Of course, the venture could adjust the money supply ex post and achieve a divide the
money pricing outcome. However, this would be simply replacing one form of commitment for
another.
Timing of payments
We now turn to consider the payments made in periods 1 and 2 of the market stage.
Discounting means that a buyer purchasing a token worth q in dollars tomorrow will only be
willing to pay ! for that token today. Therefore, for a given q, the venture will be viable if
! . To see how this works, recall that in the market stage q is known to everyone.
At the beginning of any period, there are mt tokens on issue and consumers need to purchase
tokens in order to pay the price, pt, set by the venture. By the same argument discussed before, pt
will be set to be equal to mt and the exchange rate that clears the market will be e(mt) = ntq/mt. As
a result of this, by the end of the period at least mt tokens will be held by the venture. At this
point, the venture can divest itself of those tokens. However, the willingness to pay for those
1− F(q, m1p1
)( )E[q] m1p1
E[n1q]
δq
δ (n1 +δn2 )q ≥ C
!12
tokens by others will be ! . Thus, the venture is indifferent between divesting itself of those
tokens or selling them to buyers in the next period. Regardless, the venture will earn q per period
for any period after the initial one it operates in.
The initial two periods — the ICO stage and the first period of the market stage — involve
a different timing of payments to the venture. Working backwards, if buyers hold m0 tokens at
the beginning of the market stage, they will use those tokens to purchase the product at an
exchange rate of n1q/m0. If buyers choose not to save any tokens between periods 1 and 2, at the
end of that period the venture will hold the entire supply of tokens. However, it does not receive
any influx of dollar payments during that period (i.e., period 1).
It is useful to note that if the venture holds tokens at the beginning of period 2, it always
has an incentive to release them. To see this, suppose that the venture holds a share, , of m2
(the tokens available in period 2). If it does not release any tokens, then ! . In this
case, its period 2 profits are 0. By contrast, if the venture releases those tokens, then !
and its profits are ! . The intuition is simple: the venture does not earn
any revenue in a period except by selling tokens. In the final period (i.e., period 2) this means
that even if selling tokens depreciates the currency, the venture will always find it profitable to
sell its residual holdings.
Incentives to save
In any given period, the supply of tokens is determined by several factors: the number of
tokens on issue, the number of tokens being saved for use in subsequent periods, and the number
of tokens released by the venture or others. Recall that in period 0, m0 tokens are issued. Suppose
δq
1− a
e2 = n2q / am2
e2 = n2q / m2
(1− a)e2 = (1− a)(n2q / m2 )
!13
that m1 and m2 tokens are intended to be ‘on issue’ in periods 1 and 2. A specific parameter of
interest will therefore be the growth rate in the money supply between periods 1 and 2 which we
refer to as ! (in our baseline model, we assume that ! but relax this condition in
Section 4).
To build the intuition around saving behavior, suppose that ! (i.e., the money supply
stays at a constant, m). Working backwards, let ! be the growth in demand between
periods 1 and 2. Since period 2 is the last period, the exchange rate will be:
! ,
as there is no incentive to save beyond that period. In period 1, token holders have a choice
between selling their tokens to consumers demanding access to the product in that period, or
saving them to sell them to consumers in period 2. Let s denote the share of token supply in
period 1 that is saved. Tokens will be saved so long as ! . The exchange rate in period 2 is
independent of the amount of tokens saved in period 1 while . Thus, as s rises, e1 also
rises. In equilibrium, therefore, s will rise until ! . It is easy to show that:
! .
That is, there is a positive level of saving if and only if ! . Note that all of the tokens
in the ICO are saved for at least one period. If s > 0, then ! ; while if s = 0, then
gm = m2−m1m1
gm ≥ 0
gm = 0
gn = n2−n1n1
e2 =(1+ gn )nq
m
δe2 > e1
e1 = n1q(1−s)m
δe2 = e1
s = maxδ (1+ gn )−1δ (1+ gn )
,0⎧⎨⎩⎪
⎫⎬⎭⎪
δ (1+ gn ) >1
e1 = δe2
!14
! . This means that the exchange rate will be increasing over time, while
the token-denominated price of the product will fall regardless.
What happens when the money supply changes between periods 1 and 2? In this case,
! .
If the money supply expands, this reduces the return to saving between periods 1 and 2. In
particular:
! .
If s > 0, then , which means that e1 falls as gm rises; while if s = 0, then e1 is independent
of gm.
To summarize, the incentives for token holders to save is a function of the expected growth
in demand for the product and the expected growth in the money supply. If ! ,
then s > 0, while if , s = 0. Thus, by setting gm, the entrepreneur can determine
whether saving takes place between periods 1 and 2 or not. One choice the entrepreneur has is to
set ! , in which case, ! and ! . This is the equivalent in this economy to the
Taylor Rule for monetary policy that keeps prices stable (Taylor, 1993). Note, however, that this
involves s = 0 as ! .
What about incentives to save between period 0 and period 1? First, note that there is no
demand for tokens in period 0 other than for saving purposes. Therefore, ! . This also
e1 = nqm > δ (1+ gn ) nq
m = δe2
e2 =(1+ gn )nq(1+ gm )m1
s = maxδ (1+ gn )− (1+ gm )
δ (1+ gn ),0
⎧⎨⎩⎪
⎫⎬⎭⎪
e1 = δe2
gm < δ (1+ gn )−1
gm ≥ δ (1+ gn )−1
gm = gn e1 = e2 p1 = p2
gm > δ (1+ gn )−1
e0 = δ E[e1]
!15
means that expectations regarding e1 will determine the value of the tokens that the entrepreneur
issues in period 0. Is there any reason for the entrepreneur to set ! ? Suppose the
entrepreneur does this. Then their expected return is:
!
The entrepreneur’s return is independent of m0. Thus, we can assume that in what
follows without loss in generality. Moreover, from this, we can also see that it is only the
intended gm (between periods 1 and 2) that matters and not the level of m per se.
ICO Stage
At the ICO stage, the entrepreneur cannot launch the venture unless it expects to earn
enough from the ICO to cover development costs C, i.e. unless . The value of e0 will
depend upon whether the token is expected to be saved between periods 1 and 2. Specifically,
! (when s = 0) or ! (when s > 0). Thus, the ICO will be viable and
will finance start-up costs if:
!
It is easy to see that savings choices will be made in a way to maximize e0. The entrepreneur
will, therefore, only proceed with the ICO if either of these conditions is satisfied as competition
among token purchasers will ensure that ! .
m1 > m0
e0m0 +δ E[e1](m1 − m0 ) = δ E[e1]m1 =δ E[nq]
m1m1 = δ E[nq]
δ 2 (1+gn )E[nq]m2
m1 = δ 2(1+ gn )E[nq] m1m2
if s = 0s > 0
⎧⎨⎪
⎩⎪
m ≡ m1 = m0
e0m ≥ C
e0 = δ 1m E[nq] e0 = δ 2 1+gn
m2E[nq]
δ E[nq]≥ C
δ 2 1+gn1+gm
E[nq]≥ C for
1+ gm ≥ δ (1+ gn )
1+ gm < δ (1+ gn )
e0m = max{1,δ 1+gn1+gm
}δ E[nq]
!16
It is useful to note that a venture does not gain from retaining a share (a > 0) of tokens. We
already noted that there is no return to the entrepreneur from saving tokens between periods 0
and 1. This means that the value of retaining a share of tokens arises from the ability to hold on
to them from period 0 to period 2. Recall that the entrepreneur will always want to sell any
holdings in period 2. Therefore, if a is the share of initial tokens issued (m) that are retained by
the entrepreneur, the exchange rate in period 2 will be ! , and so the return to
retaining a share a would be ! , which is only non-zero if s = 0. In that case, the return to
retaining tokens equals to , which is negative when gm is high enough so
that s = 0. What this means is that retaining a share of tokens does not perform any function than
what would otherwise be performed by setting ! , and selling new tokens into the market
(e.g. through an auction) after the ICO stage and earning seigniorage. Thus, the committed
growth in the supply of tokens is the main instrument that can impact the value of an ICO. If s >
0, then , which means that e0 falls as gm rises. If s = 0, then e0 is independent of gm.
What gm maximizes the returns to the entrepreneur? If , then s > 0 and so
the total value of tokens issued in period 0 depends (negatively) on gm. This means that the value
of the ICO is driven in part by the anticipated growth in demand for the product. By contrast, if
, s = 0, then the total value of tokens issued in period 0 is independent of gm. In
addition, the value of the ICO is independent of anticipated demand growth. Interestingly, if
! , then the value of the ICO with anticipated saving is greater than the value
e2 = (1+gn )E[nq]m(1+gm )
δ 2e2 −δe1
δ (δ (1+gn )E[nq]m(1+gm ) − E[nq]
m(1−a) )
gm > 0
e0 = δ 2E[e2]
gm < δ (1+ gn )−1
gm ≥ δ (1+ gn )−1
gm < δ (1+ gn )−1
!17
without it, and that value is falling in gm. Thus, the value of the ICO is maximized with gm=0. 11
This implies that the ICO will be viable so long as ! .
However, we have to ask whether venture profits are maximized by keeping the money
supply fixed over time. Ignoring for the moment whether the ICO value covers C or not, net of
C, the expected profits of the venture when s > 0 are:
!
By contrast, if ! , s = 0, then the expected venture profits are:
!
Thus, the outcomes with and without saving are equal and independent of gm. By setting gm = 0,
the venture can shift funds forward without changing the expected revenues from the product. Of
course, it may have other goals in mind such as price stability (which is necessary for facilitating
the use of a token as a medium of exchange). One way this can be achieved is by setting !
(thereby causing s=0). In this situation, the exchange rate and token-denominated price, p, will
be constant over time. The following proposition summarizes these results:
Proposition 2. The amount raised in an ICO is maximized by setting ! , while, conditional on raising sufficient funds to cover C, the expected net present discounted value of venture profits is independent of gm.
max{1,δ (1+ gn )}δ E[nq]≥ C
e0m +δ 2E[e2](1− s + gm )m
= δ 2 (1+ gn )E[nq](1+ gm )m
1+δ (1+ gn )( )(1+ gm )δ (1+ gn )
⎛
⎝⎜
⎞
⎠⎟ m
= δ 1+δ (1+ gn )( )E[nq]
gm ≥ δ (1+ gn )−1
δ E[nq]m
m +δ 2 (1+ gn )E[nq](1+ gm )m
(1+ gm )m = δ 1+δ (1+ gn )( )E[nq]
gm = gn
gm = 0
In Section 4 below, we show that the qualitative results of this section do not change if gm can be negative.11
!18
Comparison with equity finance
We are now in a position to compare the returns to an ICO with that of traditional equity
finance. Recall that the expected returns from an ICO are ! while the
returns from equity finance are ! . Thus, they are equivalent conditional
on each being viable. However, while equity finance is viable whenever the expected return of a
venture is positive, this does not hold for an ICO.
To see this, recall that the maximum ICO funds will be:
!
What this implies is that the ICO could involve a shortfall relative to C even if total venture gross
profits would otherwise exceed C. Total venture revenues are ! , which is the
present value of period 1 plus period 2 revenues, whereas the ICO is based on the greater of the
present values of period 1 and period 2 revenues. This arises because tokens that do not grant
their holders additional dividend, voting, and control rights do not entitle holders to a stream of
returns, but are instead ‘cashed in’ at a given point in time by investors. This issue is even more
stark when the venture has customers across more than 2 periods in the market stage (e.g. when
it plans to enter multiple industry verticals over time). For example, an infinitely-lived venture
(with no growth) will be viable if ! , whereas an ICO in which the tokens do not
constitute a crypto security will only raise! at most.
δ 1+δ (1+ gn )( )E[nq]−C
δ (1+δ (1+ gn ))E[nq]−C
e0m =δ 2(1+ gn )E[nq]
δ E[nq]
⎧⎨⎪
⎩⎪if
1< δ (1+ gn )
1≥ δ (1+ gn )
δ 1+δ (1+ gn )( )E[nq]
δn > (1−δ )C
δ E[nq]
!19
This illustrates a significant limitation of ICOs compared to equity finance that arises when
the venture is expected to be long-lived. This result is consistent with the observed use of pre-
ICO, equity-based funding rounds in this space, in which traditional VCs have funded crypto
startups before the venture and its tokens go live. The development of tokens that have similar
rights and features of traditional securities (i.e., crypto securities) may alleviate this problem,
although this also introduces new issues around the allocation of returns between token holders
and crypto equity holders which are beyond the scope of this paper.
Founders may also decide to transform, when feasible, what is typically a series of
milestone-based rounds of funding for the same venture (e.g. from Series A to IPO) into multiple,
sequential ICOs across related protocols and products. An early example of this is Protocol Labs,
which is structured as an R&D team for multiple token-based projects in the internet
infrastructure vertical. With the launch of its first project, Filecoin (which focuses on data
storage), the Protocol Labs team raised capital for only a single component of what they hope
will become a much broader, modular stack for internet services.
4. Extensions
Standard demand function
The divide the money pricing rule is simple here because there are a number of consumers
in each period, nt, each of whom have the same value for the product. A natural question to ask is
what this rule looks like when the venture faces a smooth demand function that reflects
consumers who have different willingness to pay for the product.
!20
Suppose the demand function is denoted by . In this scenario, what is the 12
optimal pricing choice for the venture at time t? If the available money supply is mt, then the
venture can target a particular nt (call it ) by choosing . At this price, for a given
exchange rate, et, the total dollar demand for tokens will be , which must
equal the dollar token supply of ! , yielding an equilibrium exchange rate of ! .
Given this, what ! will the venture target? The venture’s dollar profits are ! and, thus,
if it were to maximize the value of the tokens on issue, it will determine the quantity of
consumers at its revenue maximizing point which, in this case, is equivalent to a point where the
elasticity of demand is unity. Once again, as with Proposition 1, this is an equilibrium outcome
of the dynamic pricing game, but is not necessarily the unique outcome.
Retiring tokens
Thus far we have assumed that tokens, once issued, cannot be retired and so ! .
However, Proposition 2 shows that when the venture wants to encourage saving in order to
maximize capital raised at the ICO stage, it will want to raise the value of the tokens in the
second market period. This value depends on the amount of tokens on issue in that period. Thus,
the lower that gm is, the higher the value of the tokens in the last period. This suggests that there
might be value in committing to retire tokens between periods 1 and 2.
The challenge with this approach, however, is identifying how many tokens to retire. If the
venture committed to retire all tokens, then there would be no benefit to saving after period 1 and
nt = D(et pt )
nt* pt
* = mt
nt*
nt*et pt
* = nt*D−1(nt
*)
etmt et = nt*D−1(nt
* )mt
nt* nt
*D−1(nt*)
gm ≥ 0
Here we assume that the uncertainty regarding demand is with respect to some unknown but ex post realized 12
parameter of the demand function rather than q and n per se.
!21
hence, the ICO would be based on period 1 revenues — precisely the opposite of what is
intended by retiring tokens. Moreover, if tokens were saved, how would this retirement take
place in practice? It would look like an intentional debasing of the currency, and would require
some rationale to determine whose tokens should be retired.
Suppose that the venture could commit to retiring tokens in its possession at the end of
period 1, i.e. those used to purchase its product in period 1. If the saving rate was s, this would be
equivalent to setting . As s is endogenous and depends on gm, with this retirement
policy, the equilibrium savings rate (assuming ! ) will be:
!
In this case, ! . Thus, the value of the ICO is:
!
In this case, the maximum amount that can be raised by an ICO is less than what would arise if
the venture were to commit to gm = 0. While retirement encourages saving, that saving also
depresses the value of tokens in the second period. Furthermore, such a retiring strategy is likely
to be difficult to implement, as the venture may need those tokens to cover its operational costs,
and arbitrarily retiring tokens held by others is unlikely to be a viable option.
gm = −(1− s)
1+ gm < δ (1+ gn )
s = 1− 1− (1− s)δ (1+ gn )
⇒ s = 11+δ (1+ gn )
e2 = 1+gn(1+δ (1+gn ))m nq
e0m =δ 2 (1+gn )
1+δ (1+gn ) E[nq]
δ E[nq]
⎧⎨⎪
⎩⎪if
1< δ (1+ gn ) 1+δ (1+ gn )( )1≥ δ (1+ gn ) 1+δ (1+ gn )( )
!22
Ongoing costs
While the divide the money pricing outcome is an equilibrium that implements a positive
ongoing value for the tokens, it rests on the venture being indifferent between maximizing profits
in each period and other goals such as supplying no one or everyone. Those latter goals,
however, will cause distortions to the equilibrium value of the tokens that harm the ability of the
venture to raise funds in an ICO.
The indifference condition, furthermore, rests on there being no ongoing costs in supplying
products to token holders. Suppose that the marginal dollar cost of supplying a product to each
consumer is c > 0. For the moment, we assume that realized q exceeds c. (If it does not, the
venture will shut down even after sinking C). In this case, the venture at time t is no longer
indifferent between supplying the product to nt consumers versus another option. In particular,
since the venture has issued tokens to others in return for dollars, it receives no additional benefit
from supplying the product while it now incurs dollar costs of ! . Clearly, in this situation, the
venture would prefer to supply no products at all. Anticipating this, the value of the token at the
ICO stage (or any other stage) would be zero.
This is a potentially devastating outcome from a token value perspective. In this model, as
it involves finite time, the usual remedies to avoid such a collapse such as relational contracts do
not resolve this problem. That said, what if, at the beginning of any market stage t, but prior to
setting the token-denominated price of the product pt, the venture is required to hold a share a of
all tokens on issue? It could achieve this by either acquiring or holding on to those tokens at the
end of the previous stage, t-1. In this case, the expected dollar profit to the venture in that stage 13
cnt
After it has set its prices for that period, the venture would be free to sell those tokens into the market in order to 13
facilitate transactions. The key condition is that it holds a share, a, when it commits to a token-denominated price.
!23
would be ! . If ! , then ! and so expected dollar profits would be
! . Hence, so long as ! , the venture would choose to supply the entire market using
the “divide the money” price.
Interestingly, while by Proposition 2, conditional on being financed, the expected returns to
the venture are independent of gm, they are also unaffected by the requirement to hold a share of
tokens. However, also by Proposition 2, the requirement may limit the funds that can be raised in
an ICO, such that the venture may not be able to raise enough to cover its development costs C.
To see this, note that the maximum ICO funds would be:
!
Which simplifies to:
! .
Thus, the condition for whether the venture can proceed is:
!
By contrast, when there are ongoing costs, the feasibility condition under equity finance is:
!
In other words, even though tokens are only raised with respect to revenue drivers, the required
holding condition to prevent expropriation of token-holders means that the ICO is limited by the
profits from period one, rather than the flow of profits as in equity finance. This highlights the
aet ptnt − cnt pt = mtnt
et = ntmt
q
aqnt − cnt a ≥ cq
1− F(c,n)( )E[(1− a)e0m | q ≥ c] =1− F(c,n)( )E[(1− c
q )δ 2(1+ gn )n(q + c) | q ≥ c]
1− F(c,n)( )E[(1− cq )δn(q + c) | q ≥ c]
⎧⎨⎪
⎩⎪if
1< δ (1+ gn )
1≥ δ (1+ gn )
1− F(c,n)( )E[(1− a)e0m | q ≥ c] = 1− F(c,n)( )δ max{δ (1+ gn ),1}E[n(q − c) | q ≥ c]
δ max{δ (1+ gn ),1} 1− F(c,n)( )E[n(q − c) | q ≥ c]≥ C
δ (1+δ (1+ gn )) 1− F(c,n)( )E[n(q − c) | q ≥ c]≥ C
!24
importance of a lack of pricing commitment in constraining the amount that can be raised in an
ICO. Put simply, when there are on-going costs, the venture cannot be trusted not to expropriate
token-holders by setting a very high token-denominated price in later periods. If, using some
assumed mechanism, that commitment could be made so that the dollar denominated price was
constrained to be ! , the venture would not have to hold onto tokens at the ICO stage. In
this case, an ICO would be feasible if:
! .
This says that for q sufficiently high, the venture will set a divide the money price and, given this
commitment, it will earn the maximum (discounted) expected revenue from one of the two
market periods.
It is difficult in this general form to compare this to the feasibility condition under equity
finance. To simplify matters, suppose that n is known initially while q is uniformly distributed on
[0,1] with c < 1. In this case, ! and the amount that can be
raised in an ICO is ! . By contrast, the equity finance condition
becomes ! . Comparing the RHS of these two conditions, it can be
seen that ICO financing has a more relaxed constraint than equity financing if:
!
It is easy to see that this will hold only for c sufficiently high, while it does not hold as c gets
closer to 0. In other words, when a pricing commitment is possible and the venture does not have
max{q,c}
δ max{δ (1+ gn ),1} 1− F(c,n)( )E[nq | q ≥ c]≥ C
1− F(c)( )nE[q | q ≥ c] = n 12 (1− c2 )
δ max{δ (1+ gn ),1}n 12 (1− c2 )
δ (1+δ (1+ gn ))n 12 (1− c)2 ≥ C
max{δ (1+ gn ),1}1+δ (1+ gn )
≥ 1− c1+ c
!25
to hold a share of tokens to prevent expropriation, it potentially has an easier time raising C
upfront than would be the case under equity finance.
The question then becomes: would a venture want to raise this extra amount? Recall that
the ICO is based on the relatively lucrative market period revenues. However, the venture in this
scenario is also committed to paying for on-going costs. Thus, suppose that a venture would not
be financeable under equity finance but would be under an ICO. Then:
!
In this case, if the venture proceeds it earns:
!
This shows that with price commitments, if on-going costs are sufficiently high, an ICO can do
as well as equity finance. This arises because the ICO allows the venture to raise funds based on
the best period of revenues, mitigating the relatively low profitability of other periods. When
there are no on-going costs there is no advantage to be gained from such revenue shifting.
Third-party products and products
Thus far, we have explored a situation where the tokens are the exclusive medium of
exchange and consumers can only spend tokens to purchase products from the venture. This
represents a simplification of how the technology in used in practice, as tokens are issued to
crowdfund and bootstrap economic activity around digital ecosystems that are broader than the
venture itself. While the venture, especially in the early stages, often represents the key economic
actor within the ecosystem, its objective is to use incentives and market design to facilitate
δ max{δ (1+ gn ),1}n 12 (1− c2 ) ≥ C > δ (1+δ (1+ gn ))n 1
2 (1− c)2
δ max{δ (1+ gn ),1}n 12 (1− c2 ) +δ min{δ (1+ gn ),1}n 1
2 (1− c2 )−δ (1+δ (1+ gn ))nc(1− c)−C
= δ (1+δ (1+ gn ))n 12 (1− c)2 −C < 0
!26
transactions that extend beyond its boundaries, and create the right conditions for other
participants to join its ‘economy’. The venture could design incentives in a way to crowdsource
talent and labor (e.g. Numerai), or other key resources the ecosystem needs to scale such as
computation (e.g. Bitcoin, Ethereum), storage (Filecoin, Sia), electricity (Grid+), digital content
and data (e.g. BAT), etc. For example, in the case of a token designed to create a competitive
marketplace for data storage and data services, hard drive manufacturers and data centers can
join the digital platform developed by the venture and sell their services directly to consumers in
exchange for tokens. Third-parties can also use the shared infrastructure deployed by the venture
to develop applications on top of it that take advantage of the underlying token to settle
transactions and allocate resources.
Whereas until now we assumed that all these different types of third-parties are vertically
integrated with the venture, in this section we explicitly explore the case where they are separate
entities. To consider this, we return to the situation where there are no ongoing costs (i.e., c = 0
and the platform is purely digital) but assume that, in addition to the venture’s services, other
suppliers can accept tokens as payment in each period involving a market stage. To keep things
simple, we assume that there is a single third-party supplier who charges a token-denominated
price of wt. We assume that this supplier is the exclusive purchaser of products from the venture
and pays a token denominated price of pt to the venture to keep the platform running (e.g. so that
the venture can keep updating and maintaining the codebase). Last, to keep the model as close as
possible to our baseline, we assume that the third-party supplier’s customers place a value of q
on their product (which is initially unknown), and that there are nt such customers in each period.
!27
It is easy to see that, in each market stage, the total demand for tokens will be ! as the
third-party supplier’s customers and the supplier itself require tokens to transact on the platform,
but payment on the platform, pt, is purely internal so does not (on net) impact token demand.
Notice, however, that if the venture set its price before the third-party supplier sets its own, the
venture will set a divide the money price leaving the third-party to do the same. In this case, the
dollar demand for tokens will be ! , and the exchange rate will be the same as in the baseline
model. In other words, this structure does not change the choices the venture faces or the
conditions required for a successful financing via an ICO.
That said, if the third-party supplier has its own costs, c, it will need to earn tokens of
sufficient value to cover these costs. This will not happen if the venture sets a divide the money
price of ! . To encourage the third-party supplier, the venture could commit to receiving
a share, b, of the token payments made to the third-party, wt. In this case, the third-party will set a
divide the money price, ! , and will find it worthwhile to enter so long as ! .
Given this, the venture’s expected profits are: ! .
Note that these are decreasing in b and so will be maximized by setting b = 0. This is because, as
mentioned before, when the venture issues tokens, the initial demand for tokens is based on a
single period of token demand. So long as the third-party supplier enters, the venture will be able
to appropriate the full value of a seller for the period with the maximum discounted value of
tokens. Thus, its incentive is to ensure the third-party supplier enters for the widest feasible range
of q. This is achieved by foregoing future payments on the platform.
wtnt
qnt
pt = mt / nt
wt = mt / nt (1− b)q ≥ c
δ max{δ (1+ gn ),1}E[nq | (1− b)q ≥ c]−C
!28
Interestingly, this is related to what developers building on top of cryptocurrencies such as
Bitcoin and Ethereum refer to as “platform-level censorship resistance”: compared to traditional
digital platforms (e.g. iOS, Android etc.), platforms developed on top of crypto tokens may
provide better incentive alignment between platform architects and third-party developers of
complementary applications. While in a traditional platform the architect can expropriate ideas
and inventions from application developers and incorporate them into their offering, in a token-
based one it has no ability nor incentive to do so.
5. Imperfect commitment
The analysis so far has assumed that the venture can perfectly commit to (a) only accepting
tokens for access to the product; and (b) stating the supply of tokens upfront for each period and
not changing it under any circumstance. Whereas both commitments are fundamentally promises
by the founding team, they are typically reinforced by making the underlying codebase available
as open-source software and by hardcoding the money supply schedule within the software
protocol. Under such conditions, a fork of the network would be required to change the initial
commitments, and such fork would not be successful without the vast majority of the network
and stakeholders supporting it. We now explore the implications of relaxing the assumption that
such commitments are credible.
Money supply commitment
It is a cornerstone of monetary economics that for money to perform its function, its supply
must be tightly controlled. The same is true for crypto tokens, but there is some nuance here.
!29
To begin, suppose that while the venture can commit to m0 and m1, it cannot commit to m2.
Suppose also that between periods 1 and 2 some saving is taking place (ie., s > 0). In this case,
the period 2 venture profits would be: ! . Given that the venture still commits to
only accepting tokens in exchange for access to the platform, it uses a divide the money price of
! , while ! . Thus, its profit is ! , which is
increasing in m2. If and the venture is uncommitted to m2, then it has an incentive to set m2
as high as possible. Put simply, it has an incentive to inflate prices precisely because it does not
appropriate any returns from past saving behavior. Given this, when the commitment to m2
cannot be credibly enforced, no saving will take place.
It is useful to note that if the venture retained a share of m1, this would not remove its
incentive to expand the money supply. This is because the venture has an incentive to dispose of
such retained holdings in period 2, and these have the same profit impact as any expansion in m2.
Thus, regardless, the venture has an incentive to set m2 as high as possible if s > 0. 14
Being unable to commit to m2 can potentially impact the success of an ICO. Recall that
committing early to ! , while not changing the expected return to the venture, shifts
forward earnings so that the venture can fund C upfront. In other words, a lack of commitment
may mean some otherwise viable ventures may not be funded through an ICO. Interestingly,
when this is not a constraint, a lack of commitment on the money supply is not a problem for the
venture. While that lack of commitment means that saving will be discouraged, there is an upside
to the absence of saving behavior. If the venture holds all of the tokens, it is free to change how it
e2 p2n2 − e2sm1
p2 = m2 + sm1( ) / n2 e2 = n2q / m2 + sm1( ) qn2 −sm1
m2+sm1n2q
s > 0
m1 = m2
In practice, the use of an extremely long vesting schedule could delay this issue. However, this depends on the 14
commitment to the money supply over that time period, something that will be useful to explore in future work.
!30
operates after the period in which investors have purchased tokens to fund C, and have recouped
it by selling their token to would-be users of the product. In other words, when there is no
function performed by saving tokens, there is no value to commitment and a lack of commitment
is not an obstacle for the venture going forward.
A venture may, indeed, want to plan for such flexibility if it anticipates the need to raise
funds to finance activities that may grow the venture further. It may also find it advantageous if it
wants an exit through an IPO or acquisition that is not encumbered by previous commitments. In
summary, the conclusion here is that commitment is a cost the venture must incur in order to
shift funds forward to cover C. The less such revenue shifting is required, the less commitment is
needed. Whereas many have described ICOs as a potential substitute to traditional sources of
funding such as angel and venture capital, this highlights their complementary nature to them.
Medium of exchange commitment
The other key commitment we discussed is that the token will constitute the only way to
access the product developed by the venture. Suppose that this commitment was not maintained
and the venture, in the market stage, sold products directly to consumers in exchange for dollars.
For any buyer, the effective dollar price would be the same regardless of whether they purchased
tokens to facilitate that transaction or not. As the venture sets its token-denominated price at the
beginning of a period in the market stage, it has no incentive to set a price other than the divide
the money price. With this price, it appropriates all the consumer value which is the most it could
get by setting a dollar-denominated price instead. That said, these pricing incentives may change
if, when it sets its price, tokens are held by others outside the venture. By setting a price above
the divide the money price, no consumers would purchase tokens and the value of the tokens
!31
would depreciate (or completely collapse). In particular, this may arise if there is saving both
between periods, and between the ICO and market stage. In other words, as in the commitment
to not change the money supply, the medium of exchange commitment is critical whenever
tokens are held by others outside the venture. Imperfect commitment that allows the venture to
accept other means of payment for the product later would give the venture the incentive to set
the price so as to have payments not denominated in tokens (e.g. dollar payments) go directly to
the founding team. In this way, we can see the importance of this particular commitment for the
viability of ICOs. 15
6 . Network Effects
One of the purported benefits of ICOs is that they can assist ventures facing network effects
in avoiding coordination problems. Such problems arise when unfavorable expectations about a
network result in ventures having to use low pricing in order to generate adoption. By
comparison, a venture facing favorable expectations can price at a high level and still generate
adoption. Clearly, the more favorable the expectations for the venture and its future products, the
greater the profits for the venture. We therefore ask whether it is possible to use an ICO and
improve outcomes for ventures that would otherwise face unfavorable expectations.
To explore this, we amend the underlying demand so that the value to a consumer from the
use of the venture’s platform is ! with ! ; that is, there is a one-sided network
effect whereby the value of the platform increases as more users join. In this situation, the price
q(α + βnt ) α ,β > 0
The venture, Quantstamp, recently became embroiled in controversy when it did not adhere to a medium of 15
exchange commitment for its platform for smart contracts and was accused of accepting other cryptocurrencies and US dollars for its services (Milano and Odayar, 2018).
!32
that the venture can charge for access to the platform in a given period depends critically on
consumers’ expectations regarding how many other users will also join the platform. Using the
terminology of Hagiu (2006), if expectations are favorable, then consumers expect others to join
(i.e., ! ), and the venture can charge a price of ! in the first period. If
expectations are unfavorable instead (i.e., ! ), then the venture can only charge a price of
! . Interestingly, at that price, all consumers join and receive a surplus of ! each. Note
that there is nothing that carries over to period 2, regardless of who uses the product in period 1,
so the same coordination problem for the venture exists in this context as well.
What if expectations are unfavorable with ! ? Are such expectations sustainable if
the venture uses an ICO? The short answer is no. We are aided here by the fact that it is common
knowledge that consumers know the true value of the product q. That means that if there are bids
for the tokens that fall initially above the reserve, it is because q is sufficiently high. In that case,
those bids will come from all consumers in period 1. Thus, the total bid volume, ! , would be
based on ! which implies that the exchange rate is a perfect signal of ! .
The only equilibrium outcome in the ICO stage, therefore, is where ! .
Note that unfavorable expectations is not an equilibrium outcome as it is based on
expectations that are not fulfilled in equilibrium. Because all participants can see trading in
tokens, this allows coordination to emerge.
The same pattern can hold in period 2 where the period begins with the venture holding all
of the tokens and setting the price. The venture sets a divide the money price equal to !
E[n1] = n1 p1 = q(α + βn1)
E[n1] = 0
p1 = qα qβn1
E[n1] = 0
e0m1
δ E[n]q(α + βE[n]) E[n]
e0 = δ n1m1
q(α + βn1)
m2 / n2
!33
based on the assumption — fulfilled in any equilibrium — that if there are purchases they are
from all consumers. At this price, consumers bid for tokens with the unique equilibrium outcome
being ! . Note that if, for some reason, the venture assumed that it would have
fewer than n2 consumers — say, n — then it would set . At this price, the exchange
rate would adjust to ! . Thus, there would be rationing of customers. However, in
this model, the only rational expectations consistent with the assumption of consumer symmetry
are 0 and n2, and the venture is indifferent between pricing based on these two outcomes since it
expects 0 in either case. Therefore, the venture can set a price based on full adoption. To relate
this to the literature on network or platform pricing, note that the ICO makes the full price !
an insulating tariff (Weyl, 2010). This is a tariff that makes it a dominant strategy for each
consumer to adopt the product. That is, ! .
7. Conclusion
This paper shows that entrepreneurs have an incentive to use subsequent product pricing
choices to ensure that crypto tokens issued to fund start-up costs retain their value even when
they do not confer the typical rights associated with equity. Countering this are potential
commitment issues that arise when agents other than the entrepreneur hold tokens for any period
of time in the hope that the tokens will appreciate in value. While entrepreneurs will still price to
retain token value, they may be tempted to issue more tokens post-ICO, expropriating early
token holders. Thus, discretionary pricing is an important instrument in this context (as it allows
e2 = n2m2
q(α + βn2 )
p2 = m2 / n
e2 = nm2
q(α + βn)
et pt
et pt ≤ q(α + βnt )
!34
for price discovery), whereas discretionary monetary policy is a major concern. Such constraints
might bind if the entrepreneur needs to take advantage of expectations about future demand to
increase the value raised through an ICO and cover the development costs of a new digital
platform.
!35
References
Agrawal, A., C. Catalini, and A. Goldfarb (2013), “Some simple economics of crowdfunding,” Innovation Policy and the Economy, 14 (1), 63-97.
Athey, S., I. Parashkevov, V. Sarukkai, and J. Xia (2016), “Bitcoin pricing, adoption, and usage: Theory and evidence,” mimeo., Stanford.
Bakos, Y. and H. Halaburda (2018), The Role of Cryptographic Tokens and ICOs in Fostering Platform Adoption. Available at SSRN: https://ssrn.com/abstract=3207777 or http://dx.doi.org/10.2139/ssrn.3207777
Benedetti, H. and L. Kostovetsky (2018), “Digital Tulips? Returns to Investors in Initial Coin Offerings,” mimeo., Boston College.
Biais, B., C. Bisiere, M. Bouvard and C. Casamatta (2017), “The blockchain folk theorem,” Working Paper No.17-817, Toulouse School of Economics.
Budish, E. (2018), “The Limits of Bitcoin and the Blockchain,” mimeo., Chicago.
Canidio, A. (2018), “Financial incentives for open source development: the case of Blockchain,” mimeo., Insead.
Catalini, C. and J.S. Gans (2016), “Some simple economics of the blockchain,” mimeo, MIT https://papers.ssrn.com/sol3/papers.cfm?abstract id=2874598.
Catalini, C., J. Boslego, K. Zhang (2017), “Technological Opportunity, Bubbles and Innovation: The Dynamics of Initial Coin Offerings”, mimeo, MIT.
Chiu, J. and T.V. Koeppl (2017), “The Economics of Cryptocurrencies - Bitcoin and Beyond,” mimeo., Queens University.
Chod, J. and E. Lyandres (2018), “A Theory of ICOs: Diversification, Agency and Information Asymmetry,” mimeo., Boston College.
Cong, L.W., Y. Li and N. Wang (2018), “Tokenonomics: Dynamic Adoption and Valuation,” mimeo., Chicago.
de Bono, E. (1994), “The IBM Dollar,” Centre for the Study of Financial Innovation, London.
Fisch, C. (2018), “Initial coin offerings (ICOs) to finance new ventures: An exploratory study,” mimeo., Trier.
Hagiu, A. (2006), “Pricing and Commitment by Two-Sided Platforms,” Rand Journal of
!36
Economics, 37 (3), pp.720-737.
Halaburda, H. And M. Sarvary (2016), Beyond Bitcoin: The Economics of Digital Currencies, Palgrave MacMillan.
Howell, S.T., M. Niessner and D. Yermack (2018), “Initial Coin Offerings: Financing Growth with Cryptocurrency Token Sales,” mimeo., NYU.
Li, J. and W. Mann (2018), “Initial Coin Offering and Platform Building,” mimeo., UCLA.
Milano, A. and T. Odayar (2018), “Quantstamp under Fire: Buyers say faith shaken in $65 million token,” Coindesk (14 June); https://www.coindesk.com/quantstamp-fire-buyers-say-faith-shaken-65-million-token/
Momtaz, P.P. (2018), “Initial Coin Offerings,” Working Paper No.31, Hamburg Financial Research Center.
Nakamoto, S. (2008), “Bitcoin: A peer-to-peer electronic cash system,” mimeo, unknown.
Sockin, M. And W. Xiong (2018), “A Model of Cryptocurrencies,” mimeo., Princeton.
Taylor, J.B. (1993), “Discretion versus policy rules in practice,” Carnegie-Rochester Conference Series on Public Policy, 39, pp. 195-214.
Weyl, E.G. (2010), “A price theory of multi-sided platforms,” American Economic Review, 100 (4), pp.1642-72.
!37